Econ Writing Skill

An Agent Skill that transforms AI assistants into expert economics paper writers, synthesizing best practices from 50+ authoritative guides by Nobel laureates, Clark Medal winners, and leading economists.

What Is This?

Econ Writing Skill is an open-source Agent Skill that gives AI coding assistants deep knowledge of how to write economics papers. It distills actionable advice from over 50 guides by John Cochrane, Deirdre McCloskey, Jesse Shapiro, Keith Head, Marc Bellemare, Claudia Goldin, Edward Glaeser, Michael Kremer, and many others into a single structured skill file. Compatible with the Agent Skills open standard, it works with Claude Code and OpenAI Codex out of the box.

Key Features

Writing & Structure

• Works for all paper types -- Applied empirical, pure theory, mixed theory-empirical, structural, and descriptive papers each get tailored structure and guidance
• Dedicated formulas for every major section -- Abstract (4-part formula), Introduction (Head's formula), Model Section (Glaeser/Varian), Data Section, Conclusion (3-part formula), Title (with evaluation criteria)
• Identification strategy guidance -- Specific writing advice for 13 strategies: RCT, DiD (including staggered), IV, RDD, Synthetic Control, Synthetic DiD, Bunching, Shift-Share/Bartik, Event Studies, ML for Causal Inference, Structural Estimation, Descriptive, plus guidance for papers using multiple strategies
• Theory paper support -- Model presentation, proposition writing, and Glaeser's "start with an example" approach
• Field-specific conventions -- Macro (calibration tables, IRFs, DSGE), trade (gravity, PPML), development (CONSORT, cost-effectiveness), finance (Fama-MacBeth, winsorization)

Style & Quality

• Style rules from McCloskey and Cochrane -- Active voice, concrete language, triangular structure, reader-first writing
• Anti-AI writing patterns -- Eliminate telltale AI-generated writing (banned words, uniform sentence length, generic phrasing) and write like a real economist
• Title evaluation framework -- Score titles on clarity, specificity, length, and memorability with good/bad examples
• Citation integrity -- Anti-hallucination guidance for verifying references, distinguishing working paper vs. published versions
• Economic significance framework -- Translate coefficients into meaningful units, policy benchmarks, and back-of-envelope calculations
• Tables and figures best practices -- Self-contained captions, figure vs. table trade-offs, data visualization from Schwabish (JEP)

Review & Assessment

• Paper review and audit mode -- Simulated 3-reviewer feedback (Methodologist, Field Expert, Writing Critic) with a 100-point scoring rubric
• Pre-submission checklist -- 24-item checklist covering content, style, and modern standards
• Evaluation test suite -- 18 test cases for benchmarking skill output quality

Academic Workflow

• Job market paper (JMP) guidance -- Title, abstract, introduction polish, length, signaling breadth
• Grant proposal writing -- NSF, ERC proposal guidance: feasibility, budget justification, timeline, selling a plan vs. presenting findings
• Referee response guidance -- Point-by-point response structure for revise-and-resubmit
• Policy brief and op-ed writing -- Translate findings for non-academic audiences with plain language and concrete recommendations
• Working paper to journal conversion -- Cutting strategy, journal-specific length norms (AER, AER Insights, REStat), appendix organization
• Dissertation structure -- Three-essays format, chapter structure, authorship rules
• Survey and review paper writing -- JEL, JEP, Handbook chapter conventions
• Coauthorship conventions -- Voice consistency, designated "voice editor", signaling individual contributions

Modern Standards

• Modern empirical practices -- Pre-registration, multiple testing, specification robustness, transparency and reproducibility
• AEA replication package standards -- Data Editor requirements: README template, data citation, directory structure, code reproducibility
• AI use disclosure guidance -- Journal-specific policies (AEA, Econometric Society) for disclosing AI assistance
• LaTeX formatting guide -- Tables (booktabs, threeparttable), figures, bibliography (natbib), journal submission requirements for AER/QJE/Econometrica/REStud/JPE

Installation

Paste this into your agent (recommended)

Open Codex or Claude Code in any project and paste this once:

Install or update Econ Write as a standalone skill for the client I am using. Read and follow https://github.com/hanlulong/econ-writing-skill/blob/main/INSTALL.md. Handle same-client migration, backups, and verification yourself. Do not install for or change the other client. Do not ask me to run commands. Finish with a concise result.

The same prompt handles a first installation, future updates, and safe migration from a native plugin. It installs only for the client receiving the prompt.

Standalone commands

On macOS or Linux, install only for Claude Code:

curl -fsSL https://raw.githubusercontent.com/hanlulong/econ-writing-skill/main/scripts/install.sh | bash -s -- --global --claude

Or install only for Codex:

curl -fsSL https://raw.githubusercontent.com/hanlulong/econ-writing-skill/main/scripts/install.sh | bash -s -- --global --codex

Pass --all only when you intentionally want both clients. On Windows, or for project-local and manual installation, see Installation and updates. If the same client already has the native plugin, use the recommended agent prompt or the migration instructions so only one copy remains active.

Native plugin installation (optional)

Claude Code:

claude plugin marketplace add OpenEconAI/plugins
claude plugin install econ-write@openeconai --scope user

Explicit plugin invocation: /econ-write:econ-write.

Codex:

codex plugin marketplace add OpenEconAI/plugins
codex plugin add econ-write@openeconai

Explicit plugin invocation: $econ-write:econ-write.

Natural-language activation works with either installation method. Use one method per client; when switching methods, verify the replacement before deactivating only that client's old copy.

Usage

With the recommended standalone installation, natural-language requests can activate the skill automatically. To invoke it explicitly, use /econ-write in Claude Code or $econ-write in Codex (you can also select it from Codex's /skills menu).

Claude Code:

/econ-write write introduction for my paper on the effect of minimum wage on employment

Codex:

$econ-write rewrite this abstract to be more concrete and under 150 words

Either client can also use an ordinary request such as:

Use econ-write to draft a conclusion for my RDD paper on school funding.

Use econ-write to review this paragraph for style violations.

Use econ-write to help me structure the model section for my theory paper.

Use econ-write to audit my full paper and score it.

The skill works in four modes:

• Drafting: generates new text following the formulas and rules, marking areas that need your specific results with [AUTHOR: ...] placeholders.
• Rewriting: identifies violations of the rules (passive voice, vague language, buried leads, throat-clearing) and fixes them while preserving your meaning and contribution.
• Reviewing: checks existing text against all rules and provides a detailed report of issues with suggested fixes.
• Auditing: scores the full paper from three simulated reviewer perspectives and returns a 100-point rubric with prioritized feedback.

Common Use Cases

• Drafting a new paper section -- Generate a first draft of any section (abstract, introduction, data, results, conclusion) that follows established formulas from the start.
• Rewriting existing text for clarity -- Tighten prose, eliminate passive voice, cut throat-clearing, and make results concrete.
• Writing an introduction from scratch -- Follow Head's formula exactly: Hook, Research Question, Main Results, Literature Review & Value Added, Roadmap.
• Conducting and writing a literature review -- Build a story-driven review of the 5-10 closest papers, embedded in the introduction, that establishes your paper's niche.
• Writing up existing empirical results -- Present results in the correct order (main result first, most parsimonious to least parsimonious) with proper emphasis on economic significance.
• Writing a theory paper -- Structure model presentation, write propositions with intuition before proofs, generate testable predictions.
• Structuring a mixed theory-empirical paper -- Map empirical tests back to specific model predictions.
• Preparing a presentation -- Get advice on slide structure, pacing, and delivery following Shapiro, Cochrane, and others ("Gene Fama usually starts with 'Look at table 1.' That's a good model.").
• Reviewing a full paper -- Get a structured audit with 3 simulated reviewer perspectives and a 100-point score.
• Writing a grant proposal -- Structure an NSF, ERC, or institutional proposal that sells a research plan, not a finished paper.
• Writing for policy audiences -- Translate academic findings into plain-language policy briefs and op-eds.
• **Converting a working paper